Wednesday, October 8, 2025

AI Hacked – How Would We Know?

Since the age of sail, sailors have mastered the skill of navigation on the open seas using the stars and a sextant.  What happened when we introduced the miracle of GPS?  We completely lost our navigating skills.  Aside from a few individual throwbacks who enjoy using a sextant as a hobby, no one in the Navy can navigate without GPS.  Unbelievably, in some of the recent spate of collisions and groundings, it was discovered that bridge navigation teams had even lost the skill of fixing a position by taking bearings on known landmarks.
 
Since time immemorial, explorers have traversed the land using the stars, a map, and dead reckoning.  Our overland navigational skills increased even further with the advent of the compass.  Today, the Army has lost the ability to navigate overland without GPS.
 
Pilots used to be able to navigate cross country and hit a time on target to the second with nothing more than a map, bearings, and a stopwatch.  Today, that’s a lost skill.
 
GPS, the miracle of technology, caused us to lose our navigational skills and has made us weaker and less competent.  We have become dependent on GPS.  When our GPS systems fail or are degraded or eliminated by the enemy, we have nothing to fall back on.  Exactly like a drug addict, we have become addicted to GPS and unable to function without it.
 
What’s the next miracle of technology that we’re working so hard to acquire?  Yes, artificial intelligence (AI).  Does anyone have the slightest doubt that we’ll become utterly dependent on AI?
 
Ask a college student to write a report without using AI.  He’ll produce gibberish.  He’s lost his ability (or never developed it) to conduct research, assemble a cogent thesis, and present an intelligible, written document.  Heck, forget AI;  ask a student to write a paper without the Grammarly app and see what results.  Even simpler, ask any young adult who’s gone to public school to calculate change for a purchase in his head, without a calculator, and watch the deer in the headlights, frozen response.  We’ve become dependent on calculators and can’t even do simple arithmetic in our heads anymore.
 
Does anyone have the slightest doubt that we’ll become utterly dependent on AI?
 
The military, by the way, is attempting to make AI the foundation of our entire command and control systems.  We believe, mistakenly, that AI will give us the advantage we need to beat China.  AI, we believe, will analyze all our data, make sense of the fog of war, tell us exactly what the enemy is going to do even before he knows, and will tell us how to counter and defeat the enemy.  AI.  Magic.  One and the same.
 
Those of us who grew up during the introduction of computers are all too familiar with the well known computer programming adage, Garbage In, Garbage Out (GIGO).  Bad data in, bad results out.  AI is not immune to this phenomenon.
 
Be honest.  Does anyone seriously question what they find on the Internet?  Sure, we’ll make jokes about the Internet but does anyone actually question what they read?  Of course not.
 
Does anyone have the slightest doubt that we’ll become utterly dependent on AI?
 
Where are we going with this?  Hang in there.  We’re almost at the point of the post.  One more tidbit to assimilate.
 
Does a week go by without hearing about high level computer systems and programs, both in the military and civilian worlds, being hacked?  Nope.  And those are just the incidents that are made public.  The military and government computers and programs are hacked on a daily basis but for security reasons the incidents are kept quiet.  Despite our best efforts, various state and criminal actors routinely hack our most secure systems.  For all practical purposes, they’re unstoppable.
 
So, now put those two bits together:  absolute dependency on new technology and unstoppable hacking, and ask yourself what the result will be? 
 
The answer is easy to predict.  China will routinely hack and compromise our AI-based systems and we’ll by absolutely paralyzed because of our dependency.
 
But wait, it gets worse.  What if China hacks our AI-based systems subtly and we don’t even know it?  What if they simply manipulate the AI to give us results that give them the advantage?  We’d blindly accept the results (that’s what dependency is), never questioning them and never knowing we were being mislead and manipulated.  In fact, it would never even occur to us to ask whether the AI output was valid.
 
But wait, it gets still worse.  Even if someone was inclined to question AI results, we have no one competent enough to know what a valid result should be.  You have to have subject matter knowledge and expertise to even have an idea that something might not be right and our so-called professional warriors have no expertise (you built a ship without galvanic corrosion protection!).  So, even someone who was inclined to question a result wouldn’t have the slightest idea whether the result was or was not valid.
 
A calculator is a great tool for someone who has been trained in classical math and can recognize a garbage out result.  It is a terrible tool for someone who has no useful math skills and is unable to recognize a garbage out result.  So too, an AI command and control program could be a useful tool to a thoroughly trained and experienced professional warrior who can recognize a garbage out result.  It is a terrible tool for someone who has no useful warfighting skills and is unable to recognize a garbage out result … such as entire current flag officer corps.
 
We all recognize that networked computers are a vulnerability because if one is hacked, they’re all hacked.  We aren’t doing much to address that vulnerability but we do recognize it.  Similarly, we must recognize that AI is a vulnerability, especially when it’s being used as the basis of our highest level command and control programs.
 
Right now, just like drug dealers, we’re being given a free taste of AI to get us hooked.  We need to halt the process before we become totally addicted and helplessly dependent.  We need to regain our unaided warfighting expertise.  We do that by eliminating all non-war education (diversity, equity, gender sensitivity, climate, etc.) at the service academies, eliminating diversity crap from the leadership and ranks, ruthlessly eliminating paperwork from the daily lives of officers, eliminating deployments, bringing the fleet home for maintenance and training, start promoting a culture of acceptance of aggressiveness and ‘good’ mistakes, and start conducting daily realistic warfighting exercises and force our incompetent leadership to learn their profession.
 

14 comments:

  1. I know I'm a bad person, but at the order counter I'll hand the clerk cash, wait until they punch it in and then add some change to make the change back all paper rather than paper and coins. Typically I'm handed those extra coins back, with a very puzzled look on their face. When hiking I've never broken a map when I drop it.

    ReplyDelete
  2. I guess we could do the same thing to the Chinese and hopefully we’re already doing it.

    ReplyDelete
  3. "When our GPS systems fail or are degraded or eliminated by the enemy, we have nothing to fall back on. "

    You have, China's Beidou. Russia's Glonass, for instance, Google Pixel 10 phone has included GPS (L1+L5), GLONASS, GALILEO, BDS (Beidou)

    ReplyDelete
    Replies
    1. The hot new thing in PNT is navigating like a Pigeon,
      meaning the bird, magnetic navigation.
      SandboxAQ is one of the companies.
      https://www.sandboxaq.com/post/magnetic-navigation-revolutionizing-defense-commercial-and-civilian-navigation

      Delete
    2. Before we all burn our maps and go out and buy a magnet, let's note that there are a LOT of variables that, currently, render magnetic navigation erratic: platform's own magnetic signature (will need to calibrate for every different platform (and loadout?), altitude, temporal magnetic field variations, nearby magnetic sources (other platforms), etc. I'm aware of some test flights by the AF but I'm not aware of any routine use.

      Delete
  4. Chinese hackers are probably the least of our worries.
    The Air Force has bet the house on the F35, a less than very good aircraft, with each one containing around 400 kilograms of rare earths which we can only buy from China, while they remain sufficiently well disposed to sell them to us.
    The Navy is in a dire state, as per Conop’s innumerable posts, building aircraft carriers with non-functional catapults, and which can’t even design its own frigates, or even competently built someone else’s design.
    The Army seems to be almost entirely focused on COIN, but can’t even defeat a bunch of rice farmers/opium growers/goat herders (insert Asian peasant occupation of your choice).
    And apparently a couple dozen sea mines will collapse our entire export economy and we have no way of clearing them.
    Sounds like we’re comprehensively scre*ed
    even before things turn nasty.

    ReplyDelete
  5. Does anyone write software completely from scratch or are toolkits essential? If the toolkit is complex, it is certain to contain errors. If the toolkit has been built using toolkits, who knows how it's really working. This is before anyone deliberately introduces problems.

    You are right about the importance of classical training. We might not know the correct answer to a question we have asked ChatGPT but we can tell when it has responded with an answer which does not fit with what we already know. Unless we know very little.

    Some time ago and I forget where, I saw an image of radiation from devices being used by a small group of soldiers. To my untrained eye, they looked to be in greater danger than anyone using traditional soldiering techniques.

    I use AI quite a lot because it is so much better than me at putting together searches. It is also excellent at summarising information. Occasionally, those abilities give it the appearance of creativity but I suspect the actual intelligence of something like ChatGPT is still limited. These are early days. There will come a time when AI does much of what its boosters predict but old skills will always have a place, particularly when you cannot afford to give away your presence. Apart from anything else, the old skills are fun.

    ReplyDelete
  6. Can you imagine how a low tech (WW2 style) military could befuddle a high tech opponent. Well trained, thinking on their own with all the old skill sets. No network to hack total EMCON on the old school comms they have. (ie. saddle orders by mouth or messenger). Of course we have been dealing with this for decades now and never seem to learn anything from it.

    ReplyDelete
    Replies
    1. Supposedly, low tech is how Gen. Van Riper defeated the Navy in the Millenium Challenge 2002 war game that got everyone upset.

      Delete
  7. The only silver lining, such that it is, is that the Chinese would be in a worse situation. The US has generations of sailors. If a war broke out next year, at least there are some 60-70 year olds the USN could fall back on the emergency teach people. China does not.

    Andrew

    ReplyDelete
    Replies
    1. For some reason I don’t find that a particularly reassuring thought.

      Delete
  8. There is a way around this, and it is the same system which helped defeat Napoleon at Waterloo:

    The General Staff system.

    Naval leadership should be so well trained on what they are going to do in any conflict with the PLAN, that they need no communication to act as a team. Not Dogma, but Doctrine. Not WHAT to think, but HOW to think. Even if all data links & electronic communication are lost, each commander should be intimately familiar with how the rest of the Battle Group will *probably* fight.

    This General Staff thinking should be constantly inculcated, practiced, & wargamed, at the Naval War College. Then it should be put to the test in annual, free-play fleet problems and naval exercises.

    This was done for years before WWII. It works.

    PS: Preferably, Doctrine (How to think) should never be written down. Once it is written down, it tends to become Dogma (What to think) and becomes too inflexible. We want a set goal, but a flexible strategy to reach that goal.

    PPS: Ideally, the upper leadership of the Navy should all be initiated Disciples of this secret doctrine, with the strategic experts at the NWC (what the ancient Greeks would call the Strategos), being the High Priests. Both the practical, and the theoretical, contributing to their War Plan Yellow.

    ReplyDelete
    Replies
    1. Have to agree with this. Getting Officers back to thinking about tactics and warfare... somthing I get the feeling is barely in the top 10 of priorities these days.

      Delete
  9. "AI Hacked – How Would We Know?"

    This headline reminds me of the Stuxnet virus that we used to attack Iran's nuclear program.

    As I recall, the virus caused the centrifuges for refining nuclear material to destroy themselves. Meanwhile the readouts to the technicians showed everything to be operating as expected.

    Lutefisk

    ReplyDelete

Comments will be moderated for posts older than 7 days in order to reduce spam.